CN104992407B - A kind of image super-resolution method - Google Patents
A kind of image super-resolution method Download PDFInfo
- Publication number
- CN104992407B CN104992407B CN201510338958.6A CN201510338958A CN104992407B CN 104992407 B CN104992407 B CN 104992407B CN 201510338958 A CN201510338958 A CN 201510338958A CN 104992407 B CN104992407 B CN 104992407B
- Authority
- CN
- China
- Prior art keywords
- resolution
- mrow
- low
- msub
- msubsup
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 239000011159 matrix material Substances 0.000 claims abstract description 21
- 238000013507 mapping Methods 0.000 claims abstract description 18
- 238000005070 sampling Methods 0.000 claims abstract description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 3
- 239000012141 concentrate Substances 0.000 abstract 1
- 238000012545 processing Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000013459 approach Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Editing Of Facsimile Originals (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of image super-resolution method, comprise the following steps：The first step：Concentrate to obtain high-definition picture from view data, low-resolution image is obtained by minification down-sampling, high-resolution and low-resolution feature pair is extracted from the image pair of high-resolution and low resolution, is clustered in low resolution feature space, obtains multiple low resolution cluster centres；Second step：Using the high-resolution and low-resolution feature pair, and the low resolution cluster centre, high-resolution and low-resolution neighbour set, and mapping matrix set are calculated；3rd step：For each low resolution feature, the low resolution cluster centre the inside and its immediate cluster centre are selected, then high-resolution features are recovered with corresponding mapping matrix；The high resoluting characteristic of all reconstruct is obtained into corresponding high-definition picture block plus the image block comprising low-resolution image, all full resolution pricture blocks are fused into a full resolution pricture.
Description
Technical Field
The invention relates to the field of computer vision and image processing, in particular to an image super-resolution method.
Background
The super-resolution of the image belongs to the field of computer vision and image processing, is a classic image processing problem and has important academic and industrial research values. The aim of image super-resolution is to reconstruct a corresponding high-resolution image from a given low-resolution image, so that the visual effect is as good as possible with as little reconstruction error as possible. The current mainstream image super-resolution methods can be divided into three categories: interpolation-based methods; a reconstruction-based approach; a learning based approach.
Interpolation-based methods are a basic super-resolution method, and the processing process usually uses local covariance coefficients, fixed function kernels or adaptive structure kernels, and are widely used due to the characteristics of simplicity and rapidness. However, in many cases, the results of such methods produce visual artifacts with increasing magnification, such as: jaggies and blurring effects. Based on the reconstruction method, it is assumed that the low resolution image is obtained from the high resolution image through several degradation factors, such as: down-sampling and blurring. Such methods emphasize the importance of reconstruction constraints in the super-resolution process, and thus, the resulting high-resolution images tend to have excessively smooth and unnatural edges and ringing near the edges of the images. The learning-based approach achieves better results because a large amount of a priori knowledge is learned from the training set using machine learning techniques. However, such methods typically require that the solution be based on L0Norm or L1The processing speed of the norm optimization problem is very slow.
The above background disclosure is only for the purpose of assisting understanding of the inventive concept and technical solutions of the present invention, and does not necessarily belong to the prior art of the present patent application, and should not be used for evaluating the novelty and inventive step of the present application in the case that there is no clear evidence that the above content is disclosed at the filing date of the present patent application.
Disclosure of Invention
The present invention (main) aims to provide an image super-resolution method to solve the technical problem of slow processing speed in the prior art.
Therefore, the invention provides an image super-resolution method, which comprises the following steps: the first step is as follows: obtaining a high-resolution image from the image data set, obtaining a low-resolution image by reducing multiple and sampling, extracting high-resolution and low-resolution feature pairs from the high-resolution and low-resolution image pairs, and clustering in a low-resolution feature space to obtain a plurality of low-resolution clustering centers; the second step is that: calculating a high-resolution neighbor set, a low-resolution neighbor set and a mapping matrix set by using the high-resolution feature pair, the low-resolution feature pair and the low-resolution clustering center; the third step: for each low-resolution feature, selecting a clustering center which is closest to the low-resolution clustering center from the low-resolution clustering centers, and recovering the high-resolution feature by using a corresponding mapping matrix; and adding all the reconstructed high-resolution features to the image blocks containing the low-resolution images to obtain corresponding high-resolution image blocks, and fusing all the high-resolution image blocks into a high-resolution image.
In the third step: utilizing the statistical characteristics obtained in the first step of clustering to arrange all clustering centers in a descending order according to the element numbers; in super resolution, only the front multiple clustering centers are taken.
The clustering in the low resolution feature space in the first step comprises the steps of: obtaining a low-resolution image from the high-resolution image by using an interpolation algorithm, and then upsampling the low-resolution image set to the same size as the corresponding high-resolution image by using the interpolation algorithm; extracting a high-resolution image block set and a feature set from the high-resolution image, and extracting a low-resolution image block set and a feature set from corresponding positions in the high-resolution image; and after the low-resolution feature set is extracted, clustering the low-resolution feature set.
The second step specifically comprises: clustering the low-resolution feature pairs by using the high-resolution feature pairs and the low-resolution feature pairs to obtain a plurality of low clustering centers, and for any low-resolution clustering center, firstly finding a plurality of nearest neighbor centers, further obtaining a low-resolution feature set consisting of elements corresponding to the centers, and then finding a neighbor set of the low-resolution feature set; and after the low-resolution clustering center and the high-resolution and low-resolution neighbor sets are obtained, a mapping matrix which is mapped from the low-resolution features to the high-resolution features is further obtained by using the variables.
And in the third step, selecting the closest clustering center in the low-resolution clustering centers and obtaining the clustering center by calculating the Euclidean distance between the low-resolution features and the clustering centers.
Compared with the prior art, the invention has the advantages that: the patent application provides an image super-resolution method based on clustering and collaborative representation; by utilizing the method and the device, not only can the clustering center be determined during clustering, but also the statistical characteristics of each clustering center can be obtained, and possibility is provided for further accelerating super-resolution speed.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following detailed description and accompanying drawings. It should be emphasized that the following description is merely exemplary in nature and is not intended to limit the scope of the invention or its application.
Non-limiting and non-exclusive embodiments will be described with reference to the following figures, wherein like reference numerals refer to like parts, unless otherwise specified.
Those skilled in the art will recognize that numerous variations are possible in light of the above description, and thus the examples are intended to describe one or more specific embodiments.
The patent application provides an image super-resolution method based on clustering and collaborative representation aiming at a single low-resolution image. And extracting a training sample set from the existing high-quality image, and clustering on the training sample to obtain a low-resolution clustering center. Obtaining a high-resolution and low-resolution neighbor set corresponding to each low-resolution clustering center by considering the local geometric characteristics of data through the low-resolution clustering centers and the high-resolution and low-resolution samples, and calculating a mapping matrix from low-resolution features to high-resolution features by utilizing a collaborative representation theory, wherein the training phase is performed; for an input low-resolution image, extracting image blocks and features which are overlapped with each other, solving a low-resolution clustering center which is closest to the low-resolution features so as to obtain a corresponding mapping matrix, multiplying the mapping matrix with the low-resolution features to reconstruct high-resolution features, and adding the high-resolution image features with the low-resolution image blocks to obtain high-resolution image blocks. And finally, fusing the reconstructed image blocks together to obtain a high-resolution depth map, which is a super-resolution stage. And selecting part of clustering centers with the highest occurrence probability as a search space for super-resolution by utilizing the statistical characteristics obtained during clustering in the low-resolution feature space, thereby further accelerating the speed, which is a super-resolution accelerating process. The method not only obtains the clustering center in the low-resolution feature clustering process, but also obtains corresponding statistical characteristics, and provides a basis for later acceleration. Meanwhile, when the neighbor is organized, the local geometric characteristics of the data are considered, and the recovery effect is better. Therefore, the high-frequency information of the high-resolution image is recovered more accurately, and the high-resolution image with higher quality is obtained.
As shown in fig. 1, the present patent application provides an image super-resolution method based on clustering and collaborative representation, which includes the following steps:
a1: deriving high resolution images from natural Image common datasets (e.g. Image Net datasets)Down-sampling it to obtain a low resolution imageReducing by a multiple of s, extracting high and low resolution feature pairs from the high and low resolution image pairsAndclustering in low-resolution feature space to obtain K1Individual cluster center { ci}；
A2: by increasing the overlapping area between the image blocks and carrying out multi-scale transformation on the images, a larger number of high-resolution and low-resolution feature pairs are extractedAndand low resolution cluster center in A1 { ciCalculating high and low resolution neighbor setAndand a set of mapping matrices { F }k}；
A3: inputting a low resolution image ILExtracting low resolution image blocks therefromAnd featuresEach low resolution featureSelecting the nearest cluster center c in the low-resolution cluster centerskReuse the corresponding mapping matrix FkRecovering high resolution featuresAll reconstructed high resolution featuresAdding image blocks comprising low frequency informationObtaining corresponding high resolution image blocksFusing all high-resolution image blocks into a high-resolution image IH。
A4: and (4) utilizing the statistical characteristics obtained in the clustering in the step A1 to arrange all the clustering centers in a descending order according to the element numbers. In super resolution, only the first K' cluster centers are taken. Therefore, the search time and the storage space are saved, and the speed is further increased.
In particular embodiments, the following may be operated.
A1: for the low-resolution feature space clustering in the training stage, specifically, a) firstly using an interpolation algorithm to collect images from high resolutionObtaining a low resolution image setThen, the low-resolution image sets are up-sampled by using an interpolation algorithm until the sizes of the corresponding high-resolution images are the same; b) fromExtracting a set of high resolution image blocksAnd feature setsIn thatExtracting a set of low-resolution image blocks from corresponding positions inAnd feature setsWherein the high resolution feature yHComprises the following steps:
yH＝pH-pL, (1)
low resolution feature yLComprises the following steps:
yL＝[f1*pL；f2*pL；f3*pL；f4*pL], (2)
wherein f is1And f2Is a gradient high-pass filter in the horizontal and vertical directions, f3And f4Is a laplacian high-pass filter in the horizontal and vertical directions, and the symbol denotes the convolution operation.
Extracting a low resolution feature setThen, clustering the elements, and arranging the elements in descending order according to the number of the elements in each cluster to obtain K1Individual cluster centerThe algorithm for solving the clustering problem in b) can be K-means clustering algorithm, and the like, and the scope covered by the present patent application is not limited to the exemplified method.
A2: and calculating high and low resolution neighbor sets and a mapping matrix in a training phase. By increasing the weight between image blocksOverlapping areas and carrying out multi-scale transformation on the images, and extracting more high-resolution and low-resolution feature pairsAndfirstly, the method is carried outClustering to obtain K2Individual cluster centerFor arbitrary low resolution cluster centers ckFirst, theFinding d nearest neighbor centers to obtain a low resolution feature set composed of elements corresponding to the d centersThen according to the following rulesFind its max neighbor set NL.i：
Where max is a threshold for controlling NL.iNumber of middle elements, NN (Nearest Neighbor) represents Nearest Neighbor, max NN representsMiddle distance cluster center ckThe nearest NN neighbors. The symbol | | | represents the number of elements in the set. At a corresponding positionExtracting to form a high-resolution neighbor set NH.i。
After obtaining the low-resolution clustering center and the high-resolution and low-resolution neighbor sets, assuming a low-resolution feature yLThe low resolution cluster center and nearest neighbor set are ckAnd NL,kTo obtain reconstructed high resolution features yHThe required coefficient x, the following optimization objective function is solved first:
wherein,representing solving x to minimize the value of the objective function;represents L2The square of the norm; λ is a parameter greater than zero; the problem has an analytic solution, and the concrete form is as follows:
wherein (C)TMeans for representing a matrix, ()-1Representing the inverse of the matrix; the corresponding high-resolution features can then be determined by the following equation:
most of the above equation is y with inputLIrrelevant, and therefore can be calculated off-line, i.e. as the mapping matrix:
thus, for each low resolution cluster center, its corresponding mapping matrix can be found. Specifically, the optimization problem may be solved by a collaborative expression method, and the scope covered by the present patent application is not limited to the exemplified method.
A3: in the super-resolution stage, for an arbitrarily input low-resolution image ILExtracting low resolution image blocks therefromAnd featuresFor each low resolution featureSelecting the nearest cluster center c in the low-resolution cluster centerskReuse the corresponding mapping matrix FkRecovering high resolution featuresAll reconstructed high resolution featuresAdding image blocks comprising low frequency informationObtaining corresponding high resolution image blocksFusing all high-resolution image blocks into a high-resolution image IH. Specifically, the distance between the low-resolution feature and the cluster center may be calculated by euclidean distance, and the scope covered by the present application is not limited to the illustrated method.
A4: in the super-resolution acceleration stage. Using the statistical characteristics obtained in the clustering in a1, namely: the number of elements in each cluster is different. And (4) sorting all the clustering centers in a descending order according to the number of the elements. In super resolution, only the first K' cluster centers are taken, and although the number of the cluster centers is small, the sum of corresponding elements is very large. For each input low-resolution feature, only the nearest neighbor center needs to be searched from the K' clustering centers, so that the search time and the storage space are saved, and the speed is further increased.
The patent application provides an image super-resolution method based on clustering and collaborative representation. Solving with the present patent application is based on L2The norm optimization model accelerates the speed, the learned statistical characteristics are utilized to reduce the search area in the super-resolution process, the speed is further accelerated, and when the nearest neighbor is organized, the high-frequency information of the high-resolution image is recovered more accurately by considering the local geometric characteristics of the data, so that the high-resolution image with higher quality is obtained.
The patent application provides an image super-resolution method based on clustering and collaborative representation; by utilizing the method and the device, not only can the clustering center be determined during clustering, but also the statistical characteristics of each clustering center can be obtained, possibility is provided for further accelerating super-resolution speed, and simultaneously, the collaborative expression theory is utilized, so that most of calculation can be performed off-line, the super-resolution speed is accelerated, and when the neighbor is organized, the local geometric characteristics of data are utilized for clustering the search space, so that the high-frequency information of the high-resolution image is recovered more accurately, and the high-resolution image with higher quality is obtained.
While there has been described and illustrated what are considered to be example embodiments of the present invention, it will be understood by those skilled in the art that various changes and substitutions may be made therein without departing from the spirit of the invention. In addition, many modifications may be made to adapt a particular situation to the teachings of the present invention without departing from the central concept described herein. Therefore, it is intended that the invention not be limited to the particular embodiments disclosed, but that the invention will include all embodiments and equivalents falling within the scope of the invention.
Claims (4)
1. An image super-resolution method is characterized by comprising the following steps:
the first step is as follows: obtaining a high-resolution image from the image data set, obtaining a low-resolution image by reducing multiple and sampling, extracting high-resolution and low-resolution feature pairs from the high-resolution and low-resolution image pairs, and clustering in a low-resolution feature space to obtain a plurality of low-resolution clustering centers;
the second step is that: calculating a high-resolution neighbor set, a low-resolution neighbor set and a mapping matrix set by using the high-resolution feature pair, the low-resolution feature pair and the low-resolution clustering center;
the second step specifically comprises: by increasing the overlapping area between the image blocks and carrying out multi-scale transformation on the images, a larger number of high-resolution and low-resolution feature pairs are extractedAndfirstly, the method is carried outClustering to obtain K2Individual cluster centerFor arbitrary low resolution cluster centers ckFirst, theFinding d nearest neighbor centers to obtain a low resolution feature set composed of elements corresponding to the d centersThen according to the following rulesFind its max neighbor set NL.i：
<mrow> <msub> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msubsup> <mi>Y</mi> <mrow> <mi>L</mi> <mi>w</mi> </mrow> <mi>i</mi> </msubsup> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>i</mi> <mi>f</mi> <mi>m</mi> <mi>a</mi> <mi>x</mi> <mo>&GreaterEqual;</mo> <mo>|</mo> <msubsup> <mi>Y</mi> <mrow> <mi>L</mi> <mi>w</mi> </mrow> <mi>i</mi> </msubsup> <mo>|</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mi>max</mi> <mi>N</mi> <mi>N</mi> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>i</mi> <mi>f</mi> <mi>m</mi> <mi>a</mi> <mi>x</mi> <mo><</mo> <mo>|</mo> <msubsup> <mi>Y</mi> <mrow> <mi>L</mi> <mi>w</mi> </mrow> <mi>i</mi> </msubsup> <mo>|</mo> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>,</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </mrow>
Where max is a threshold for controlling NL.iNumber of middle elements, NN for nearest neighbor, max NN forMiddle distance cluster center ckThe nearest NN neighbors; the symbol | | represents the number of elements in the set; at a corresponding positionExtracting to form a high-resolution neighbor set NH.i；
After obtaining the low-resolution clustering center and the high-resolution and low-resolution neighbor sets, assuming a low-resolution feature yLThe low resolution cluster center and nearest neighbor set are ckAnd NL,kTo obtain reconstructed high resolution features yHThe required coefficient x, the following optimization objective function is solved first:
<mrow> <munder> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> <mi>x</mi> </munder> <mo>|</mo> <mo>|</mo> <msub> <mi>y</mi> <mi>L</mi> </msub> <mo>-</mo> <msub> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mi>x</mi> <mo>|</mo> <msubsup> <mo>|</mo> <mn>2</mn> <mn>2</mn> </msubsup> <mo>+</mo> <mi>&lambda;</mi> <mo>|</mo> <mo>|</mo> <mi>x</mi> <mo>|</mo> <msubsup> <mo>|</mo> <mn>2</mn> <mn>2</mn> </msubsup> <mo>,</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> </mrow>
wherein,representing solving x to minimize the value of the objective function;represents L2The square of the norm; λ is a parameter greater than zero; the analytic solution is:
<mrow> <mi>x</mi> <mo>=</mo> <msup> <mrow> <mo>(</mo> <msubsup> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> <mi>T</mi> </msubsup> <msub> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>+</mo> <mi>&lambda;</mi> <mi>I</mi> <mo>)</mo> </mrow> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msup> <msubsup> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> <mi>T</mi> </msubsup> <msub> <mi>y</mi> <mi>L</mi> </msub> <mo>.</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> </mrow>
wherein (C)TMeans for representing a matrix, ()-1Representing the inverse of the matrix; the corresponding high-resolution features are obtained by the following formula:
<mrow> <msub> <mi>y</mi> <mi>H</mi> </msub> <mo>=</mo> <msub> <mi>N</mi> <mrow> <mi>H</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mi>x</mi> <mo>=</mo> <msub> <mi>N</mi> <mrow> <mi>H</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <msup> <mrow> <mo>(</mo> <msubsup> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> <mi>T</mi> </msubsup> <msub> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>+</mo> <mi>&lambda;</mi> <mi>I</mi> <mo>)</mo> </mrow> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msup> <msubsup> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> <mi>T</mi> </msubsup> <msub> <mi>y</mi> <mi>L</mi> </msub> <mo>,</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>6</mn> <mo>)</mo> </mrow> </mrow>
the mapping matrix is:
<mrow> <msub> <mi>F</mi> <mi>k</mi> </msub> <mo>=</mo> <msub> <mi>N</mi> <mrow> <mi>H</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <msup> <mrow> <mo>(</mo> <msubsup> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> <mi>T</mi> </msubsup> <msub> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>+</mo> <mi>&lambda;</mi> <mi>I</mi> <mo>)</mo> </mrow> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msup> <msubsup> <mi>N</mi> <mrow> <mi>L</mi> <mo>,</mo> <mi>k</mi> </mrow> <mi>T</mi> </msubsup> <mo>.</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>7</mn> <mo>)</mo> </mrow> </mrow>
for each low-resolution clustering center, obtaining a corresponding mapping matrix;
the third step: for each low-resolution feature, selecting a clustering center which is closest to the low-resolution clustering center from the low-resolution clustering centers, and recovering the high-resolution feature by using a corresponding mapping matrix; and adding all the reconstructed high-resolution features to the image blocks containing the low-resolution images to obtain corresponding high-resolution image blocks, and fusing all the high-resolution image blocks into a high-resolution image.
2. The image super-resolution method according to claim 1, characterized in that: in the third step: utilizing the statistical characteristics obtained in the first step of clustering to arrange all clustering centers in a descending order according to the element numbers; in super resolution, only the front multiple clustering centers are taken.
3. The image super-resolution method according to claim 1, characterized in that: the clustering in the low resolution feature space in the first step comprises the steps of: obtaining a low-resolution image from the high-resolution image by using an interpolation algorithm, and then upsampling the low-resolution image set to the same size as the corresponding high-resolution image by using the interpolation algorithm; extracting a high-resolution image block set and a feature set from the high-resolution image, and extracting a low-resolution image block set and a feature set from corresponding positions in the high-resolution image; and after the low-resolution feature set is extracted, clustering the low-resolution feature set.
4. The image super-resolution method according to claim 1, characterized in that: and in the third step, selecting the closest clustering center in the low-resolution clustering centers and obtaining the clustering center by calculating the Euclidean distance between the low-resolution features and the clustering centers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510338958.6A CN104992407B (en) | 2015-06-17 | 2015-06-17 | A kind of image super-resolution method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510338958.6A CN104992407B (en) | 2015-06-17 | 2015-06-17 | A kind of image super-resolution method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104992407A CN104992407A (en) | 2015-10-21 |
| CN104992407B true CN104992407B (en) | 2018-03-16 |
Family
ID=54304215
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510338958.6A Active CN104992407B (en) | 2015-06-17 | 2015-06-17 | A kind of image super-resolution method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104992407B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105787899A (en) * | 2016-03-03 | 2016-07-20 | 河海大学 | Rapid image super-resolution method based on self-adaptive regression |
| CN106339984B (en) * | 2016-08-27 | 2019-09-13 | 中国石油大学(华东) | Distributed image ultra-resolution method based on K mean value driving convolutional neural networks |
| CN106327428B (en) * | 2016-08-31 | 2019-12-10 | 深圳大学 | image super-resolution method and system based on transfer learning |
| CN110084752B (en) * | 2019-05-06 | 2023-04-21 | 电子科技大学 | Image super-resolution reconstruction method based on edge direction and K-means clustering |
| CN115082319B (en) * | 2022-07-22 | 2022-11-25 | 平安银行股份有限公司 | Super-resolution image construction method, computer equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101976435A (en) * | 2010-10-07 | 2011-02-16 | 西安电子科技大学 | Combination learning super-resolution method based on dual constraint |
| CN102831581A (en) * | 2012-07-27 | 2012-12-19 | 中山大学 | Method for reconstructing super-resolution image |
| CN102902961A (en) * | 2012-09-21 | 2013-01-30 | 武汉大学 | Face super-resolution processing method based on K neighbor sparse coding average value constraint |
| CN103530863A (en) * | 2013-10-30 | 2014-01-22 | 广东威创视讯科技股份有限公司 | Multistage reconstruction image super resolution method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101874482B1 (en) * | 2012-10-16 | 2018-07-05 | 삼성전자주식회사 | Apparatus and method of reconstructing 3-dimension super-resolution image from depth image |
-
2015
- 2015-06-17 CN CN201510338958.6A patent/CN104992407B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101976435A (en) * | 2010-10-07 | 2011-02-16 | 西安电子科技大学 | Combination learning super-resolution method based on dual constraint |
| CN102831581A (en) * | 2012-07-27 | 2012-12-19 | 中山大学 | Method for reconstructing super-resolution image |
| CN102902961A (en) * | 2012-09-21 | 2013-01-30 | 武汉大学 | Face super-resolution processing method based on K neighbor sparse coding average value constraint |
| CN103530863A (en) * | 2013-10-30 | 2014-01-22 | 广东威创视讯科技股份有限公司 | Multistage reconstruction image super resolution method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104992407A (en) | 2015-10-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104899830B (en) | A kind of image super-resolution method | |
| CN110120011B (en) | Video super-resolution method based on convolutional neural network and mixed resolution | |
| CN104992407B (en) | A kind of image super-resolution method | |
| CN110136062B (en) | Super-resolution reconstruction method combining semantic segmentation | |
| CN103984959B (en) | A kind of image classification method based on data and task-driven | |
| CN105488776B (en) | Super-resolution image reconstruction method and device | |
| US11301965B2 (en) | Method and image processing device for image super resolution, image enhancement, and convolutional neural network model training | |
| CN103761710B (en) | The blind deblurring method of efficient image based on edge self-adaption | |
| CN106339984B (en) | Distributed image ultra-resolution method based on K mean value driving convolutional neural networks | |
| CN109727195B (en) | Image super-resolution reconstruction method | |
| CN102402784B (en) | Human face image super-resolution method based on nearest feature line manifold learning | |
| CN111861886B (en) | Image super-resolution reconstruction method based on multi-scale feedback network | |
| CN106530231B (en) | Super-resolution image reconstruction method and system based on deep cooperative expression | |
| Muqeet et al. | Hybrid residual attention network for single image super resolution | |
| WO2015180055A1 (en) | Super-resolution image reconstruction method and apparatus based on classified dictionary database | |
| CN115953784A (en) | Laser coding character segmentation method based on residual error and feature blocking attention | |
| CN113658206B (en) | Plant leaf segmentation method | |
| CN114037720A (en) | Pathological image segmentation and classification method and device based on semi-supervised learning | |
| CN115867933A (en) | Computer-implemented method, computer program product and system for processing images | |
| CN116895037A (en) | Frame insertion method and system based on edge information and multi-scale cross fusion network | |
| Behjati et al. | Hierarchical residual attention network for single image super-resolution | |
| CN115393868B (en) | Text detection method, device, electronic equipment and storage medium | |
| CN111681168A (en) | Low-resolution cell super-resolution reconstruction method based on parallel residual error network | |
| CN104867107B (en) | A kind of image super-resolution method | |
| CN104123707A (en) | Local rank priori based single-image super-resolution reconstruction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |